Jan 2023 structural break in visit counts in Advan Monthly Pattern

I’m analyzing monthly raw visit counts for several NAICS using the Advan Monthly Pattern. Across many geographies, I see a large, systematic drop (often ~40–60%) from Nov/Dec 2022 to Jan 2023. The pattern shows up broadly, not just in a few regions. I’ve tested multiple normalizations/scalings methods (e.g., by device, population, number of POIs), but the Jan ’23 discontinuity persists.

Did the provider implement changes around Jan 2023, e.g., device panel size/weighting, visit inference/dedup, privacy thresholds, outlier filters, or POI taxonomy/reclassification? Thanks!